Research, innovations and leaders need to support using artificial intelligence (AI) in electronic health record (EHR) systems to encourage widespread adoption and take the advancing tech into mainstream healthcare.
Traditional, manual and antiquated electronic health systems are undergoing a digital transformation. Artificial intelligence (AI) is increasingly entering healthcare, transforming existing online information by removing the cumbersome, complex and confusing nature of retaining and maintaining electronic health records (EHRs), sometimes known as electronic medical records (EMRs).
AI-led EHR systems have the potential to contribute to democratising access to informed, accurate and timely care. However, innovators must explore the tech鈥檚 capabilities to provide equitable healthcare and understand how it can meet individuals’ and communities’ specific needs.
Tech to transform EHR capabilities
Health technology pioneers are learning how AI-led discoveries can help provide this healthcare standard. They are responding by exploring how technological features in EHR systems can create a more high-performing digital health ecosystem.
鈥淎I has the potential to transform current EHRs from being passive information storage systems that organise health data into active information-generating systems that surface actionable clinical insights from health data,鈥� says Steven Lin, Executive Director of the (HEA3RT) at Stanford University School of Medicine.
Using AI, clinicians can make previously hidden patterns and insights in patent data visible. In turn, unlocking this data can lead to substantial improvements in a patient鈥檚 treatment and wider population care. 鈥淓HRs, initially built to store and allow the retrieval of patients鈥� health data, are shifted into a far more functional realm by AI,鈥� says Peter Fish, CEO of Mendelian.
US Tariffs are shifting - will you react or anticipate?
Don鈥檛 let policy changes catch you off guard. Stay proactive with real-time data and expert analysis.
By GlobalDataAs the advancing tech grows in the face-to-face clinical setting, practitioners change how they store and access health records. 鈥淒esigned thoughtfully, an AI-powered EHR can become the physician’s most powerful tool and even a trusted partner,鈥� says Lin.
However, concerns exist around utilising AI鈥檚 capabilities in EHR, preventing the tech from entering mainstream healthcare. 鈥淒esigned poorly, it can obfuscate and interfere with patient care and worsen the epidemic of physician burnout,鈥� Lin adds.
Predictive algorithms shape personalisation
AI-led EHR systems are accelerating to deliver personalised healthcare. AI in EHR aims to give practitioners a proactive tool for personalised healthcare management of chronic and vulnerable patients.
鈥淎s medicine moves from the one-size-fits-all approach into stratified and ultimately personalised medicine, it becomes more and more complex to deliver the care patients deserve at scale,鈥� says Fish.
Predictive algorithms integrated into EHRs can support clinical decision-making. AI-embedded tools can predict whether a patient will have a certain percentage chance of being hospitalised for a particular condition and will recommend appropriate and responsive intervention.
The tech captures computational pattern-matching capabilities which exceed practitioners鈥� own at this level, Fish says. Predictive algorithms can also improve the experience of using EHRs for physicians by personalising menus, buttons, layouts and shortcuts to the physician’s pattern of use.
New developments and emerging research
鈥淚’m excited about AI-driven risk prediction in primary care and population health,鈥� says Lin. Identifying patients at high risk of preventable outcomes such as heart attacks, strokes, emergency department visits and hospitalisations and intervening using evidence-based recommendations can save lives and money, Lin says.
鈥淚’m also excited about the next generation of AI-assisted clinical decision-making tools that can help physicians make the best treatment plans for 鈥榩atients like theirs鈥�,鈥� Lin adds. Phenotyping patients and using real-world evidence to personalise care are examples of how we can expect AI to develop amid ongoing research.
Mendelian, a rare disease diagnostics healthcare company, has launched MendelScan, a clinical decision support system. It is designed to allow and deploy rare disease case-finding algorithms on EHRs asynchronously, as it aims to form a population-level proactive care process.
Translating tech into care for people鈥檚 personalised needs
鈥淚t would be disingenuous to say that AI in EHRs is having a measurable impact on the care or personalisation of care on actual patients right now,鈥� says Lin. Despite 鈥減ockets of success鈥�, Lin adds that healthcare does not exhibit anything 鈥渂road or meaningful yet鈥� in the AI-led EHR space. 鈥淭here is real potential, but the proof points are not there,鈥� Lin adds.
Significant challenges exist in the sphere, specifically in maximising AI鈥檚 capabilities in EHR systems and providing opportunities to grow their potential and improve overall global healthcare.
Many successful stories of AI in EHRs have yet to be used at scale. There are more stories of embarrassing failures, such as than there are successes, Lin shares. Poor predictions and demonstrations of AI have widespread and potentially long-standing effects. 鈥淸It] significantly impacts physician and patient trust in AI, which is low overall,鈥� says Lin.
鈥淎nother challenge is the general unwillingness and resistance of large EHR vendors to work with third-party AI developers, stifling innovation and progress,鈥� continues Lin. Aligning legacy healthcare systems and processes with innovations featuring new technology is an ongoing obstacle. Therefore, partnerships between third-party AI developers and large EHR providers remain largely unexplored.
鈥淭he continued lack of interoperability between EHRs remains a major barrier,鈥� Lin adds. Disparate systems that do not communicate with one another are a limitation that stifles progression for practitioners looking for AI to improve their existing information-gathering processes.
鈥淲e see a lot of complexity around data sharing that can be solved technically,鈥� says Fish. With careful planning, the EHR data is incomplete and sometimes inaccurate, and driving large-scale adoption proves problematic. 鈥淚t may take years for national commissioners to decide on the next steps,鈥� relays Fish.
Future AI potential in EHR systems
鈥淎I holds huge amounts of promise for medicine,鈥� Fish details. Yet, significant real-world pilots and effectiveness studies are required to generate the robust evidence required to entice decision-makers and gatekeepers. Today, we need leaders to champion and support AI in EHR systems to enable their trust and widespread adoption.
A significant gap in our understanding is how best to integrate AI models into human-driven clinical workflows. 鈥淲e need to invest significantly more into the implementation science of AI,鈥� says Lin.
Understanding the real-life capabilities of AI-led tech is one of many demands. Ensuring it is seamless and appealing is critical too. 鈥淚t doesn’t matter how good the AI is, if the technology causes too much friction on existing human-driven workflows, it will not be adopted,鈥� Lin highlights.
